New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Feature]: Allow logging artifact without creating new versions #7413
Comments
Related: #6278 |
Hi @buoyancy99, thank you for writing in! I have submitted your feature request to our enginering team. |
+1, this is a huge problem !!! |
If you are using pytorch lightning, I have a work around here (could be useful for the team too)
|
Description
Now wandb has tightened its storage policy, we found it's very space-inefficient to store all versions of artifacts. For example, to save space, if I am to store model weights I may only want to store the latest version. However, I found no good ways to overwrite artifact with the same name without creating versions.
Note: pytorch_lightning's ModelCheckpoint class has options like
save_top_k
orenable_version_counter
, but these only affect the behavior for local files, not wandb. Wandb creates a new version anyway so it wastes a lot of storage.Suggested Solution
Add an
overwrite
option torun.log_artifact
such that whenoverwrite=True
, log_artifact will not create new version but instead overwrite the artifact with same name if it existsAlternatives
No response
Additional Context
No response
The text was updated successfully, but these errors were encountered: